etcd

etcd is a distributed key-value store which is highly available, strongly consistent, and watchable for changes. The name "etcd" was from a unix's configuration directory, "etc" and "d"istributed system. etcd is adopted by cloud-native systems such as Kubernetes, Cloud Foundry Diego, and Project Calico. Major uses cases include metadata store and distributed coordination.

History

Originally etcd was started for two use cases: reboot coordination and application configuration. CoreOS used etcd to coordinate reboot of CoreOS cluster and avoid that all nodes in the cluster rebooted at the same time. Also, etcd was used to store application configuration, so whenever a server starts or restarts or whenever application configuration is updated, the server receives application configuration from etcd.

Query Compilation

Not Supported

Data Model

Key/Value

etcd stores data as a multiversion key-value pair. Each mutative operation (e.g. PUT) creates a new version and does not change older versions. Previous versions are also accessible until they are compacted.

Indexes

B+Tree

etcd creates an in-memory btree index for keys and provides range operations.

System Architecture

Shared-Nothing

The etcd cluster is composed of shared-nothing nodes. The cluster has one leader node and other nodes work as followers, which will be determined at run-time (Raft algorithm). When the leader node receives a request, the leader takes votes against all followers. If the majority of nodes agrees on the request, the leader commits the request and ask followers to commit. An etcd client does not need to know which node is a leader to send a request because a follower node can forward the request to the leader node.

Storage Architecture

Disk-oriented

etcd stores a key-value pair in a persistent disk as the b+ tree structure sorted by a key.

Logging

Command Logging

The documentation mentioned that etcd used WAL to recover from a failure but did not mention what kind of information it logs. According to the source code, it seems to log a gRPC request in WAL.

Storage Model

N-ary Storage Model (Row/Record)

etcd stores physically data as a key-value pair. The key is consist of a 3-tuple: major, sub, type. Major contains the revision (a counter which is incremented when data modification is requested.) Sub contains the identifier among the revision because the transaction might produce a single revision with multiple keys. Type is an optional and one use case is for a tombstone. The value contains a delta from a previous version.

Stored Procedures

Not Supported